Zurich Open Repository andArchiveUniversity of ZurichMain LibraryStrickhofstrasse 39CH-8057 Zurichwww.zora.uzh.ch

Year: 2018

Cell type-specific analysis of transcriptome changes in the porcineendometrium on Day 12 of pregnancy

Zeng, Shuqin ; Bick, Jochen ; Ulbrich, Susanne E ; Bauersachs, Stefan

Abstract: BACKGROUND: Along with trophoblast elongation (Days 10 to 12), estradiol is secreted inincreasing amounts for recognition of pregnancy. Endometrial secretions driven by ovarian progesteroneand conceptus signals are essential for conceptus growth and development. Results of transcriptome anal-yses of whole endometrial tissue samples in the pig indicated the need for cell type-specific endometrialgene expression analysis for a better understanding of transcriptome changes associated with establish-ment of pregnancy. RESULTS: The most distinct transcriptome profile and the majority of differentiallyexpressed genes (DEGs) were identified in luminal epithelium (LE). Many DEGs were found only in thecell type-specific analysis. The functional classification of DEGs identified in specific endometrial celltypes revealed various distinct functions and pathways. Genes related to immune activation, estrogensignalling pathway, embryo development, and cell proliferation were upregulated in LE of pregnant gilts.Genes involved in sterol biosynthetic and metabolic processes and extracellular matrix were upregulatedin stroma. Genes associated with cell communication such as via exosomes and vesicles were found asdifferential in LE, glandular epithelium (GE), and stroma (S). CONCLUSIONS: This study revealed thatconceptus signals induce different transcriptomic regulations in the endometrial compartments/cell typesrelated to their specific function during recognition and establishment of pregnancy.

DOI: https://doi.org/10.1186/s12864-018-4855-y

Posted at the Zurich Open Repository and Archive, University of ZurichZORA URL: https://doi.org/10.5167/uzh-152925Journal ArticlePublished Version

The following work is licensed under a Creative Commons: Attribution 4.0 International (CC BY 4.0)License.

Originally published at:Zeng, Shuqin; Bick, Jochen; Ulbrich, Susanne E; Bauersachs, Stefan (2018). Cell type-specific analysis oftranscriptome changes in the porcine endometrium on Day 12 of pregnancy. BMC Genomics, 19(1):459.DOI: https://doi.org/10.1186/s12864-018-4855-y

RESEARCH ARTICLE Open Access

Cell type-specific analysis of transcriptomechanges in the porcine endometrium onDay 12 of pregnancyShuqin Zeng1,2, Jochen Bick1, Susanne E. Ulbrich1 and Stefan Bauersachs2*

Abstract

Background: Along with trophoblast elongation (Days 10 to 12), estradiol is secreted in increasing amounts forrecognition of pregnancy. Endometrial secretions driven by ovarian progesterone and conceptus signals areessential for conceptus growth and development. Results of transcriptome analyses of whole endometrial tissuesamples in the pig indicated the need for cell type-specific endometrial gene expression analysis for a betterunderstanding of transcriptome changes associated with establishment of pregnancy.

Results: The most distinct transcriptome profile and the majority of differentially expressed genes (DEGs) wereidentified in luminal epithelium (LE). Many DEGs were found only in the cell type-specific analysis. The functionalclassification of DEGs identified in specific endometrial cell types revealed various distinct functions and pathways.Genes related to immune activation, estrogen signaling pathway, embryo development, and cell proliferation wereupregulated in LE of pregnant gilts. Genes involved in sterol biosynthetic and metabolic processes and extracellularmatrix were upregulated in stroma. Genes associated with cell communication such as via exosomes and vesicleswere found as differential in LE, glandular epithelium (GE), and stroma (S).

Conclusions: This study revealed that conceptus signals induce different transcriptomic regulations in theendometrial compartments/cell types related to their specific function during recognition and establishment ofpregnancy.

Keywords: Pig, Pregnancy, Estrous cycle, Uterus, Endometrium, Transcriptomics, RNA-Seq

BackgroundUterine receptivity to conceptus implantation beginsafter hatching from the zona pellucida. The embryopre-contacts with uterine luminal (LE) or superficialglandular (sGE) epithelia; later followed by appositionand adhesion between trophectoderm and uterine LE/sGE. Finally, limited or extensive endometrial invasion(depending on species and type of placenta) initiates pla-centation. These peri-implantation events are prerequi-sites for fetal and placental growth and developmentthrough the remainder of pregnancy [1, 2]. During theconception cycle, the porcine embryo undergoes the firstcell divisions in the oviduct and arrives in the uterus in

the 4 cells stage [3]. After hatching from the zona pellu-cida on Day 7, the porcine conceptus changes rapidlyfrom spherical to tubular, and then filamentous formbetween Days 10 and 12 after ovulation [4]. This isaccompanied by differentiation of trophoblast cells mak-ing the conceptus ready for implantation [5]. Maternalrecognition and establishment of pregnancy in pigsrequires a biphasic pattern of estrogen secretion by theconceptus, mainly 17β-estradiol (E2), increased betweenDays 11–12 and Day 15 and Days 25–30, respectively[6]. Previous studies also indicated that the porcine em-bryo exerts local effects on endometrial structures andfunctional parameters as well [7, 8]. The effects on theendometrial luminal epithelium structure, uterine bloodflow [9], histotroph production and changes in prosta-glandin F2α (PGF2a) release are mainly caused by E2through directly or indirectly mediating the release ofother substances [10]. Influences on cellular organization,

* Correspondence: stefan.bauersachs@uzh.ch2Department for Farm Animals, University of Zurich, Genetics and FunctionalGenomics, Clinic of Reproductive Medicine, Zurich, SwitzerlandFull list of author information is available at the end of the article

© The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, andreproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link tothe Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Zeng et al. BMC Genomics (2018) 19:459 https://doi.org/10.1186/s12864-018-4855-y

composition, and function derived from E2 effects isthought to facilitate in turn conceptus development andlater plancentation [11]. The pro-inflammatory cytokineinterleukin one beta 2 (IL1B2) which expression is dra-matically increased only in expanding porcine conceptuses(Day 11–12), is proposed to contribute to embryoimplantation [12]. Interleukin one beta 2 binds to theinterleukin 1 receptor type 2 (IL1R2) on the epithelial sur-face which can activate the nuclear factor kappa-B (NFKB)system [13], followed by cytokines (tumor necrosis factor(TNFα), Interleukins (ILs), interleukin 6 family cytokine(LIF), and colony stimulating factor 2 (GMCSF)), chemo-kines (C-C motif chemokine ligand 5 (RANTES)), and thePTGS2 signaling pathway [14].The embryo-endometrial cross-talk in the pig has

evolved as a critical mechanism responding initially tothe hormonal trigger in order to achieve receptivity andthen amplify the decision under embryonic signals forpregnancy. Thus, the endometrium’s receptive abilityand the prevention of luteolysis is very important duringthis critical period [5]. So far, a number of studies havebeen performed to analyze differential gene expressionin porcine endometrium in response to embryonic sig-nals starting from blastocyst stage [15] until placentation[16–19]. In our previous studies we investigated differ-ential gene expression in porcine endometrium on Days12 and 14 of pregnancy by the use of RNA-sequencing(RNA-seq) and identified numerous DEGs in responseto the presence of conceptuses [18, 19]. Previous studiesin the mouse conducted with Laser capture microdissec-tion (LCM), a commonly used method to isolate differ-ent cell types from whole tissues or organs with highpurity [20, 21], and microarrays indicated celltype-specific differences in transcriptome changes be-tween LE and GE in response to pregnancy and revealedthat different cell types act differentially and synergistic-ally to permit blastocyst implantation [22]. Field et al.studied the immune-related gene expression in endo-metrial epithelial and stromal cells by LCM and microar-rays, and found that the expression of factors involved inimmunomodulation/endometrial remodeling was differ-ent between the epithelial and stromal compartments ina temporal manner [23]. Although the mouse is a goodmodel to study the uterine transcriptome, the resultscan not be directly transferred to other species. Previ-ously, Brooks et al. focused on the ovine cell-specificuterine transcriptome by the use of LCM and found anumber of genes that contribute to cell proliferation,migration, attachment, and differentiation in ovine uter-ine LE [24]. Although a number of studies using LCM todissect the endometrial cell types have been performed,none of them compared more than two cell types anddid a comparison to the complete tissue sample. Sincethe knowledge about the cell-type specific gene

expression is rather limited in porcine endometrium, theaim of the present study was to characterize the complextranscriptome changes in different endometrial cell typesduring the preimplantation period for improving ourunderstanding of localization of endometrial geneexpression regulation in context of recognition of preg-nancy. To do this analysis in a transcriptome-wide man-ner, RNA-seq was performed for samples derived fromluminal epithelium (LE), glandular epithelium (GE), andstroma (S) isolated by the use of LCM and compared toa corresponding data set for complete endometrial tissuesamples.

MethodsIsolation of target cellsThe animal experiment and sample collection is de-scribed in Samborski et al. [18]. Treatments of gilts wereperformed in accordance with the regulations of thelocal authorities (District Government of Upper Bavaria,Veterinary Office). The performed standard procedures/treatments in animal breeding were all in accordancewith the International Guiding Principles for Biomed-ical Research Involving Animals, as proposed by theSociety for the Study of Reproduction, with the Euro-pean Convention on Animal Experimentation andwith the German Animal Welfare Act. The animalswere housed at farm facilities of the LMU Munich(Chair for Molecular Animal Breeding and Biotech-nology) and slaughtered at the slaughterhouse of theInstitute for Animal Breeding, Bavarian State ResearchCenter for Agriculture, Poing, Germany.A number of 8 prepuberal gilts were bought from a

livestock trader (crossbreeds of German Landrace andPiétrain) and received a single injection of 750 IU PMSG(Intergonan®, MSD Animal Health Innovation GmbH,Schwabenheim, Germany) and 72 h later 750 IU hCG(Ovogest®, MSD Animal Health Innovation GmbH) tosynchronize ovulation. Gilts of the “pregnant” group (n =4) were inseminated twice (24 h and 36 h after hCG) witha standard dose of German Landrace semen whereas giltsof the “non-pregnant” control group (n = 4) were insemi-nated with the supernatant of centrifuged (10 min,3000 rpm) semen of the same boar. Gilts were slaughteredon Day 12 after insemination. Endometrial samples (med-ial part of the uterine horns) were collected andsnap-frozen in liquid nitrogen and stored at − 80° untilpreparation for LCM. Pregnancy was confirmed by thepresence of filamentous conceptuses in the flush of theuterine horns. Briefly, frozen endometrium samples werecut in 10 μm thick sections with a Leica CM1950 clinicalcryostat (Leica Biosystems, Germany), mounted ontomembrane slides (MembraneSlide NF 1.0 PEN, Zeiss,Germany), and stained using a modified, rapid Cresyl vio-let staining protocol to identify LE, GE, and stromal cells.

Zeng et al. BMC Genomics (2018) 19:459 Page 2 of 19

Briefly, the slides were first fixed with 70% ethanol(Sigma), and quickly washed in 50% ethanol. Cresyl violetwas used to stain for 3 min, and then the stained slideswere washed again with 50, 70, 100% ethanol, respectively(dip slides two/three times into each solution). Finally, theslides were dried at room temperature. All solutionswere prepared with RNase-free water. The target cellsfrom the sections were captured using a LCM Zeiss200 M (inverse) microscope (Zeiss PALM Microsys-tems, Germany). When satisfactory cutting wasachieved, the target tissue was lifted to the LCM cap(AdhesiveCap 200 clear, Zeiss, Germany) and 50 μlextraction buffer was used to incubate the LCM sam-ples at 42 °C for 30 min to lyse the cells.

Isolation of RNA, quality control, and RNA-sequencingTotal RNA was isolated from luminal epithelium, glan-dular epithelium, and stromal samples of each pig usingthe PicoPure RNA Isolation Kit (Applied Biosystems™,Vilnius, Lithuania) following the manufacturer’s instruc-tions. Integrity and quantity of the RNA were assessedusing the Agilent RNA 6000 Pico assay on the Agilent2100 Bioanalyzer (Agilent Technologies, Waldbronn,Germany). The quality of the isolated total RNAextracted from LE, GE, and S were ranging from 6.6 to9.4 (RNA integrity number, RIN). Most of the sampleshad a RIN around 8 were used to prepare a total of 24RNA-seq libraries for biological replicates per group(n = 4 gilts) and cell type.The Ovation SoLo Single Cell RNA-Seq System

(NuGen Technologies, San Carlos, USA) was used forpreparing RNA-Seq libraries starting from 500 pg oftotal RNA (corresponding to no more than 50 cells) ac-cording to the manufacturer’s recommendations. Thenumber of PCR cycles for the first amplification step(determined by qPCR according to the manual) was be-tween 15 and 20, for most of the samples 17. The 24 li-braries were prepared from each individual sample forthe three cell types of each replicate of both experimen-tal groups. Then, all individual barcoded libraries werepooled for sequencing on two lanes of a single-read flowcell on an Illumina HiSeq 2500 instrument. Sequencingand demultiplexing were performed at the FunctionalGenomics Center Zurich (FGCZ).

Bioinformatics and data analysisThe obtained sequence reads (Fastq files) were analyzedwith a locally installed version of Galaxy [25]. Theadapter sequence was clipped (for shorter fragmentswhere sequencing runs into the adapter) and 5 bp wereremoved from the start of the reads. Sequences weremapped with Hisat2 (Sscrofa 11.1) from NCBI (ftp://ftp.ncbi.nih.gov/genomes/Sus_scrofa/GFF) and filteredby CPM cutoff after removing the PCR duplicates with

NUGEN nudup based on the 8 random bases containedin the barcode adapter. The resulting read count tablewas used for statistical analysis in EdgeR to identify dif-ferentially expressed genes (DEGs) [26]. With a false dis-covery rate (FDR) of < 1% for LE and using itscorresponding p value of 0.0021 as a cut-off for GE,stroma, and complete tissue, hierarchical cluster analysiswas performed for DEGs in MultiExperiment Viewer(MeV). Gene ontology (GO) and pathway analysis wasconducted by using the online tool ToppCluster and theDatabase for Annotation, Visualization, and IntegratedDiscovery (DAVID) [27]. The data set from completeendometrial tissue samples was analyzed with the samepipeline except the step for removal of PCR duplicates.Raw FASTQ files were deposited at National Center forBiotechnology Information (NCBI) Gene ExpressionOmnibus (GSE109539).

ResultsTranscriptome sequencing of samples collected by LCMCells from uterine LE, GE, and S were isolated from en-dometria collected from Day 12 pregnant gilts (n = 4)and Day 12 nonpregnant cyclic gilts (control, n = 4)using LCM. Although the isolated stromal areas com-prise a mix of fibroblasts and other cell types, mainlydifferent immune cells, stroma (S) is in the following re-ferred to as a “cell type”. A total of 458.1 million rawreads were obtained from the two experimental groups;205.4 and 252.7 million in pregnant and control groups,respectively. After low-quality reads were filtered outand PCR duplicates were removed, a total of 259.3 mil-lion clean reads (116.1 in pregnant group, 143.2 millionin control group) were selected for further analyses(Additional file 1: Table S1).

Detected genes and DEGs in comparison of completetissue and LCM samplesThe obtained LCM RNA-seq expression data were com-pared to data from complete endometrium tissue sam-ples collected from Day 12 pregnant gilts andcorresponding cyclic controls (raw data from Samborskiet al. [18]). A total of 12,401 genes were detectable withthe LCM method. For the individual endometrial com-partments these were 11,415, 11,676, and 11,532 genes(LE, GE, and S, respectively). Clearly more genes(16260) were detectable in complete endometrial tissuesamples. A high number of genes (10576) could be de-tected in the three LCM sample types and in completetissue (Fig. 1a). Only 170 genes were detectable in theLCM samples but not in the complete sample. Thenumbers of genes putatively showing cell type-specificexpression were 317, 344, and 161, for LE, GE, andstroma, respectively (expression detected in the corre-sponding cell type and for most of the genes also in

Zeng et al. BMC Genomics (2018) 19:459 Page 3 of 19

complete tissue). Eleven solute carrier family members(SLCs) and 5 genes coding for transmembrane protein(TMEMs) were only detected in LE. In addition, 10coiled-coil domain containing (CCDC) mRNAs wereonly found in GE. These genes detected only in one celltype were mostly expressed at rather low levels (data notshown).The analysis of differential gene expression was per-

formed between the pregnant and the nonpregnantgroup for each cell type and also for complete tissue

samples. More DEGs were found for LE when comparedwith GE and S, i.e., 2388, 246, and 597 DEGs, respect-ively (Fig. 1b, Additional file 2: Table S2). Almost 5000DEGs were obtained for the complete tissue samples incomparison of pregnant and nonpregnant gilts (Fig. 1b,Additional file 2:Table S2). Many of the DEGs, mainlyfor LE with 1920 DEGs, were differentially expressed(DE) in one cell type but not in the other cell types or incomplete tissue (Fig. 1b, sector overlap LE withComplete and sector LE only). Also, for GE and S, manyDEGs were specifically DE in these cell types (Fig. 1b).About half of all genes found to be DE in a celltype-specific way were not found as DE in the completetissue sample.

Unsupervised clustering of samples by multipledimension scaling plotsMultiple dimension scaling (MDS) plot analysis revealedfor principal component 1 a grouping according to theindividual cell types, particularly for the LE in the preg-nant state (Fig. 2a). The GE and S samples derived frompregnant endometria were more similar to each otherthan the corresponding samples derived from the cyclicgroup. In the second dimension (principal component2), a clear separation of pregnant and control sampleswas mainly found for LE, but also for GE and S (Fig. 2a).The lowest distance between the pregnant and cycliccontrol group was observed for GE corresponding to thelowest number of DEGs.Furthermore, hierarchical cluster analysis was performed

for the DEGs identified for each cell type (Figs. 2b, c, d).With respect to higher or lower expression in pregnantcompared to cyclic endometrium, 1260 up-regulated and1128 down-regulated, 193 up-regulated and 53down-regulated, and 231 up-regulated and 366down-regulated DEGs were identified in LE, GE and S, re-spectively (Additional file 2: Table S2).

Functional category analysis of DEGsAn overview of the network illustrating functional classi-fication of the obtained DEGs is shown in Fig. 3. AllDEGs of LCM were analyzed with ToppCluster tool forGene Ontology (GO) and pathway analysis. Selected spe-cifically enriched functional categories are shown forbiological process (BP), molecular function (MF), cellu-lar component (CC), and pathways for LE, GE, and S,respectively. The majority of overrepresented categoriessuch as adhesion junction, cell motility, embryo develop-ment and MAPK cascade were obtained for LE. Cellularhomeostasis, metal transport, and steroid metabolicprocess were found for stroma. Peptidase inhibitor activ-ity and membrane transporter activity were enriched forGE samples. A systematic comparison of overrepresentedGO categories and KEGG pathways between the

Fig. 1 Venn diagram showing the overlaps of detectable genes (a)and differentially expressed genes (b). Green, luminal epithelium (LE);blue, glandular epithelium (GE); pink, stromal cells (S); yellow,complete tissue samples

Zeng et al. BMC Genomics (2018) 19:459 Page 4 of 19

endometrial cell types and complete endometria was per-formed with the DAVID tool. At a FDR of 1%, 47 overrep-resented functional categories and KEGG pathways werefound in total. For selected categories, the FDR and thefold enrichment obtained for the three cell types andcomplete endometria was used to generate a heatmap inorder to illustrate specific and common overrepresentedcategories and pathways (Figs. 4a and b). Based on this

comparison, DEGs were enriched for the functional cat-egories “extracellular exosome” and “extracellular vesicle”in all three cell types as well as complete tissue. Specificenrichment of DEGs in LE was found for categories re-lated to cell communication and signaling (e.g. estrogensignaling), immune response, epithelium development,cell proliferation, cytoskeleton, cell junction and migra-tion. In contrast, DEGs found in stroma were enriched forfunctional categories and pathways involved in sterol andsteroid biosynthesis and metabolism, inflammatoryresponse, extracellular matrix, and ion transport. Particu-lar enrichment for DEGs in complete endometria wasidentified for categories related to cell division and bloodvessel morphogenesis. For GE, the number of enrichedterms and pathways was generally very low due to the lownumber of DEGs. More detailed information about thecomparison of overrepresented categories and pathwaysand the assigned DEGs for specific cell types and completetissue is shown in Additional file 3: Table S3.

Genes differentially expressed in specific cell types or incomplete tissueFirst, the log2 fold changes (FC) (pregnant/control) ofthe top 10 up- and downregulated DEGs for each celltype and complete endometrial tissue were compared(Fig. 5, Additional file 4: Table S4). Genes such asmultimerin 1 (MMRN1), prospero homeobox 1(PROX1), sodium voltage-gated channel alpha subunit3 (SCN3A), synuclein alpha (SNCA), sclerostin do-main containing 1 (SOSTDC1), LOC102165712 wereonly DE in stroma. The genes neuropeptide Y (NPY),osteocrin (OSTN), hemoglobin subunit epsilon 1(HBE1), LOC106510525, LOC110257993 could only befound as DE in complete tissue. Some members ofthe S100 calcium binding protein A family werehighly upregulated in all three cell types, except S100calcium binding protein A7 (S100A7) which was spe-cifically upregulated in LE. Furthermore, serpin familyB members 2 and 7 were highly upregulated in allcell types whereas serpin family B member 11 (SER-PINB11) in LE only (not detectable in GE andstroma).Genes were defined as specifically DE if they were

contained in the Venn diagram in the not overlappingsectors of each cell type and the sector for the over-lap of one cell type with the DEGs found in thecomplete tissue samples (Fig. 1b). Furthermore, geneswere analyzed that were DE only in the complete tis-sue samples, which may well contain genes DE in celltypes not present in the LCM samples. Functional an-notation results (overrepresentation analysis) for genesonly DE in certain cell types or complete tissue areshown in Additional file 5: Table S5. The main over-represented functional categories for the genes

Fig. 2 Unsupervised clustering of endometrial LCM samples. a. Amultidimensional scaling plot (principal component analysis) for the500 genes showing the highest pairwise fold changes between thesamples in the dataset for LCM samples was performed in EdgeR.Samples from the same group are shown in the same color. b-dHierarchical cluster analysis of differentially expressed genesidentified for the luminal epithelium (b), glandular epithelium (c),and stroma (d). Mean-centered expression values (log2 counts permillion of sample – mean of log2 counts per million of all samples)for the samples of the control and pregnant groups are shown forgenes with significant differences in gene expression (FDR < 1%).Color scale is from − 2 (blue, lower than mean) to 2 (red, higherthan mean). Each row represents 1 gene, each column 1 sample.The first letter of the sample IDs indicates the group (C or P: controlor pregnant group) and the second the cell type (L, G, and S:luminal epithelium, glandular epithelium, and the stromal cells),followed by the ID of the animal, respectively

Zeng et al. BMC Genomics (2018) 19:459 Page 5 of 19

specifically upregulated in LE were related to “phos-phorus metabolic process”, “ATP binding”, “signaltransduction”, “cell adhesion/junction”, “cell migra-tion”, “extracellular vesicle”, “amide and peptide bio-synthetic process”, “protein transport”, “cytoskeletonorganization”, “immune response”, and “cell cycle”.The genes specifically downregulated in LE were in-volved in “metal ion binding”, “cell morphogenesis”,“epithelium development”, “microtubule cytoskeleton”,and “DNA repair”. Only a few overrepresented func-tional categories were found for GE-specific DEGs,“endoplasmic reticulum” and “extracellular vesicle” forupregulated genes. Specific DEGs for stromal regionswere enriched for “ion transport”, “cytoskeletonorganization”, “cell morphogenesis”, and “blood vessel

development” (upregulated genes) and “cellular lipidmetabolic process”, “steroid biosynthetic process”,“endoplasmic reticulum”, and “metal ion binding”(downregulated genes). Highly significant enrichmentwas obtained for many functional categories in theanalysis of upregulated genes specific for completetissue. The main processes and functions were “cellcycle”, “cell proliferation”, “vasculature or blood vesseldevelopment”, “cell motility”, “cell death”, “proteoly-sis”, “signal transduction”, “cell adhesion”, “leukocyteactivation”, and “extracellular vesicle”. For the down-regulated genes, specific for complete tissue, overrep-resentation was found for “mitochondrion”, “steroiddehydrogenase activity”, “ciliary part”, “ciliumorganization”, and a variety of metabolic processes.

Fig. 3 Gene Ontology (GO) functional classification network. All significant genes (human Entrez Gene IDs) in three cell types were used as inputfor the ToppCluster. The following databases were used, i.e. “biological process”, “cellular component”, “molecular function” and pathway. Finally,the data were uploaded in Cytoscape 3.4.0 to modify the network. Nodes were colored based on specificity: red nodes specific for luminalepithelium (LE), glandular epithelium (GE), and stromal cells (S); nodes for the three GO functions and pathway were in different colors

Zeng et al. BMC Genomics (2018) 19:459 Page 6 of 19

DEGs involved in estrogen signaling, prostaglandinmetabolism and signaling, and other selected pathwaysA number of 249 DEGs assigned to a selection ofparticularly interesting pathways and processes, e.g.,estrogen signaling, steroid hormone biosynthesis, prosta-glandin (PG) metabolism, signaling and transport,interleukin-1 (IL1) and interferon (IFN) type I signal-ing are shown in Additional file 6: Table S6. A totalnumber of 122 DEGs including 81 up- and 41 down-regulated genes were found as DE in LE. For the GE,16 were up- and 3 downregulated. In addition, anumber of 20 genes were upregulated and 22 geneswere downregulated in stroma. For complete endo-metrial samples, 117 assigned genes were upregulatedand 69 were downregulated. For some of the path-ways, functionally related genes showed distinct

localization of gene expression regulation and in partthe same direction of regulation (i.e., collective up- ordownregulation).

Comparison of the LCM RNA-seq results to data fromreal-time RT-PCR and published localization studiesOverall, the overlap of the DEGs between the LCM sam-ples and the complete tissue samples (dataset from Sam-borski et al. [18]) showed the reliability of the LCMRNA-seq data. Furthermore, a comparison to results ofquantitative PCR for 25 selected genes from Samborskiet al. [18] and other studies was performed (Table 1).Furthermore, the localization of endometrial gene ex-pression of 20 selected genes observed by LCMRNA-seq was compared to results from in situhybridization (ISH) and immunohistochemistry (IHC)

S GE LE Complete LE Complete GE Sa b

Fig. 4 Comparison of significance of enrichment (false discovery rate, FDR) of selected functional categories and pathways between endometrialcompartments and complete endometrium samples (a). Overrepresented functional categories and pathways were selected (FDR < 1%) from theresults obtained for each cell type and the complete endometria. The FDR from 0 (black) to 1 (white) is shown for selected categories/pathwaysspecific or in common for the studied cell types. Each row represents a function or pathway, each column a cell type or complete tissue (luminalepithelium (LE), glandular epithelium (GE), stromal cells (S), and complete tissue). Comparison of fold enrichment of selected functional categoriesand pathways between endometrial compartments and complete endometrium samples (b). Overrepresented functional categories andpathways were selected (FDR < 1%) from the results obtained for each cell type and the complete endometria. The fold enrichment from 1(white, no enrichment) to 3 (red, 3-fold enrichment) is shown for selected categories/pathways specific or in common for the studied cell types.Each row represents a function or pathway, each column 1 cell type or complete tissue (luminal epithelium (LE), glandular epithelium (GE),stromal cells (S), and complete tissue)

Zeng et al. BMC Genomics (2018) 19:459 Page 7 of 19

(Table 2). Both comparisons showed very good agree-ment and in addition the superior sensitivity in compari-son to ISH and IHC.

DiscussionThe present study implemented laser capture microdis-section (LCM) to isolate endometrial LE, GE, andstroma from the porcine uterus with the aim of uncover-ing important cell-specific gene expression regulation,which is masked in the transcriptome analysis of wholeendometrium samples due to the complex cell typecomposition of this mucosal tissue. Overall, the obtainedresults and the clear sample clustering based on theRNA-seq data generated by this approach agreed verywell with our previous studies utilizing RNA-seq ofcomplete tissue samples and the confirmation of se-lected DEGs by real-time RT-PCR [18, 19]. Furthermore,a comparison to published results for localization of ex-pression of selected genes in porcine endometrium

revealed good agreement and showed higher sensitivityof the LCM RNA-seq approach compared to ISH andIHC. In comparison to the data derived from completeendometrial tissue, less detectable genes were found forthe certain cell types/compartments. This may be attrib-uted to the fact that the endometrial tissue has a com-plex and dynamic cellular composition, which includesLE, GE, fibroblasts, endothelial cells, pericytes, and vari-ous immune cells. However, the number of more than11,500 detectable genes obtained from the analysis ofthe LCM samples is in the expected range of transcribedgenes for the analysis of an individual cell type.The results for the LCM samples revealed that there

was a clear difference in gene expression betweennon-pregnant and pregnant samples, particularly for theLE. This indicates that the main effects of the conceptussignaling on endometrial gene expression are localizedto the luminal epithelium. Furthermore, more than halfof the genes DE in LCM samples were not found as

Fig. 5 Heatmap for the log2 fold changes (pregnant/control) of the top 10 differentially expressed genes of each gene expression comparison.The scale is from log2 FC -7 (blue, downregulated in pregnant samples) to log2 FC 7 (red, upregulated in pregnant samples). Each row represents1 gene, each column 1 cell type or tissue (luminal epithelium (LE), glandular epithelium (GE), stromal cells (S), and complete tissue)

Zeng et al. BMC Genomics (2018) 19:459 Page 8 of 19

Table

1Com

parison

ofRN

A-seq

andqP

CRdata

Sscge

nesymbo

lSscEntrez

Gen

eID

Hsa

gene

symbo

l

Hsa

Entrez

Gen

eID

Hsa

gene

descrip

tion

log2

FCP/C

P-value*

RNA-seq

qPCR

RNA-seq

qPCR

LEGE

SCom

plete

D12

D14

LEGE

SCom

plete

D12

D14

PubM

edID

ABCC

1733,619

ABCC

14363

ATP

bind

ingcassette

subfam

ilyCmem

ber1

3.5

1.4

1.1

2.4

1.6

–< 0.001

0.224

0.321

<0.001

< 0.01

–27764917

LOC100738425

100,738,425

ABCC

410,257

multid

rugresistance-associatedprotein4

nd2.9

nd2.2

2.1

––

0.019

–0.001

< 0.001

–24695625

ABCC

9100,127,449

ABCC

910,060

ATP

bind

ingcassette

subfam

ilyCmem

ber9

1.2

0.0

0.6

2.2

1.8

–< 0.001

0.999

0.844

<0.001

< 0.05

–27764917

AKR1B1

396,816

AKR1B1

231

aldo

-ketoredu

ctasefamily

1mem

berB

5.2

5.2

2.2

5.1

5.8

–< 0.001

< 0.001

0.060

<0.001

< 0.001

–24695626

BAALC

100,170,128

BAALC

79,870

brainandacuteleukem

ia,cytop

lasm

icnd

ndnd

−5.0

−5.8

––

––

<0.001

< 0.001

–24174570

LOC100157834

100,157,834

CDH17

1015

cadh

erin

17,LIcadhe

rin(liver-intestine)

−6.5

−2.6

−5.8

−8.3

−6.3

–< 0.001

0.001

< 0.001

<0.001

< 0.001

–24174570

CLDN1

100,625,166

CLDN1

9076

claudin1

5.0

nd4.2

3.4

1.8

–< 0.001

–< 0.001

<0.001

0.002

–25123632

PBD-2

404,699

DEFB1

1672

defensin,b

eta1

−5.2

−1.7

−4.1

−6.4

−5.9

−3.4

< 0.001

0.015

< 0.001

<0.001

< 0.001

0.010

24174570

FGF9

396,717

FGF9

2254

fibroblastgrow

thfactor

9(glia-activating

factor)

2.2

ndnd

3.0

3.7

3.0

< 0.001

––

<0.001

< 0.001

< 0.001

24174570

FGFR3

100,514,115

FGFR3

2261

fibroblastgrow

thfactor

receptor

3−1.2

−0.1

−1.1

−2.3

–− 0.6

< 0.001

0.991

0.066

<0.001

–0.048

20393170

LOC100511475

100,511,475

FXYD

453,828

FXYD

domaincontaining

iontransport

regu

lator4

−4.9

nd−5.2

−7.4

−7.5

–< 0.001

–< 0.001

<0.001

< 0.001

–24174570

LOC100513220

100,513,220

GPR83

10,888

Gprotein-coup

ledreceptor

83−5.9

nd−7.5

−7.8

−4.0

–< 0.001

–< 0.001

<0.001

< 0.001

–24174570

HPG

D100,156,186

HPG

D3248

hydroxyprostagland

inde

hydrog

enase15-

(NAD)

−3.8

−1.4

−3.6

−3.0

−2.7

–< 0.001

0.005

< 0.001

<0.001

< 0.01

–24695626

LOC100739374

100,739,374

IL24

11,009

interleukin

24−2.8

−1.4

− 2.7

−4.6

−5.2

−5.3

< 0.001

0.254

< 0.001

<0.001

0.024

0.033

24174570

IRF1

396,611

IRF1

3659

interfe

ronregu

latory

factor

10.6

−0.1

0.4

0.4

0.5

2.9

0.115

0.890

0.397

0.087

0.188

0.003

24174570

IRG1

100,524,951

IRG1

730,249

immun

orespo

nsive1ho

molog

(mou

se)

10.3

5.8

6.2

9.9

8.0

–< 0.001

0.000

< 0.001

<0.001

< 0.001

–24174570

OSTN

100,049,691

OSTN

344,901

osteocrin

ndnd

nd10.3

12.4

––

––

<0.001

< 0.001

–24174570

S100A9

100,127,489

S100A9

6280

S100

calcium

bind

ingproteinA9

12.4

11.5

11.3

10.4

10.6

9.2

< 0.001

< 0.001

< 0.001

<0.001

< 0.001

< 0.001

24174570

Zeng et al. BMC Genomics (2018) 19:459 Page 9 of 19

Table

1Com

parison

ofRN

A-seq

andqP

CRdata

(Con

tinued)

Sscge

nesymbo

lSscEntrez

Gen

eID

Hsa

gene

symbo

l

Hsa

Entrez

Gen

eID

Hsa

gene

descrip

tion

log2

FCP/C

P-value*

RNA-seq

qPCR

RNA-seq

qPCR

SERPINB7

100,152,588

SERPINB7

8710

serpin

peptidaseinhibitor,cladeB,mem

ber7

10.4

7.9

6.2

10.3

8.9

5.4

< 0.001

< 0.001

< 0.001

<0.001

< 0.001

0.018

24174570

SLCO

2A1

100,144,510

SLCO

2A1

6578

solute

carrierorganicaniontransporterfamily

mem

ber2A

11.0

0.4

2.3

0.5

3.0

–0.011

0.960

< 0.001

0.053

< 0.001

–24695625

SLCO

4C1

100,737,557

SLCO

4C1

353,189

solute

carrierorganicaniontransporterfamily

mem

ber4C

11.6

0.2

1.3

2.3

2.6

–0.007

0.978

0.174

<0.001

< 0.05

–27764917

SLCO

5A1

100,157,250

SLCO

5A1

81,796

solute

carrierorganicaniontransporterfamily

mem

ber5A

18.4

4.9

5.9

6.4

9.2

–< 0.001

< 0.001

< 0.001

<0.001

< 0.05

–27764917

SPP1

397,087

SPP1

6696

secreted

phosph

oprotein

14.1

1.6

3.2

2.4

2.5

3.2

< 0.001

0.005

< 0.001

<0.001

0.008

0.029

24174570

LOC100738308

100,738,308

STAT1

6772

sign

altransducer

andactivator

oftranscrip

tion

1,91

kDa

1.6

0.2

0.8

1.4

1.2

2.3

< 0.001

0.562

0.066

<0.001

0.001

< 0.001

24174570

STC1

100,125,345

STC1

6781

stanniocalcin1

−2.9

−1.0

−2.9

1.9

2.5

2.0

< 0.001

0.384

< 0.001

<0.001

0.101

0.017

24174570

Ssc:Susscrofa;H

sa:H

omosapiens;P/C:p

regn

antvs.con

trol;*forRN

A-seq

adjusted

p-values;n

d:no

tde

tectab

le

Zeng et al. BMC Genomics (2018) 19:459 Page 10 of 19

differential in complete tissue, showing that analysis ofthe whole endometrium is partially missing celltype-specific gene regulation. Some genes not appearingas DE in whole endometria can even be expressed bothin LE and GE or stroma but with opposite expressionregulation, which eventually results in similar expres-sion between pregnant and non-pregnant state incomplete tissue.

Genes preferentially differentially expressed inendometrial cell types are related to specific functionsFor the genes specifically DE in LE overrepresentedfunctional categories were mainly related to signaling,immune functions, and cell adhesion. This is probablyreflecting the various responses to the signals derivedfrom the elongating conceptus mainly affecting the LE.Genes upregulated in LE and assigned, e.g., to “proteinphosphorylation” and other categories related to signal-ing could be induced by conceptus-derived interleukin 1beta which has been suggested to play an important rolein development of the LE by stimulating cell prolifera-tion via activation of the ERK1/2 and P38 MAPK cellsignaling cascades [28]. Furthermore, endometrial- and/or conceptus derived epidermal growth factor (EGF) hasbeen shown to stimulate phosphorylation of ERK1/2

MAPK in porcine LE cells and proposed to regulatedevelopment of the peri-implantation uterine LE at thefetal-maternal interface [29].For stroma, DEGs, related to “vasculature develop-

ment”, “ion transport”, “cell development”, were overrep-resented. Angiopoietin-2 (ANGPT2) mRNA wasdetectable in all three cell types but only upregulated instroma and complete endometria in this study. ANGPTscomprise a second key group of vascular regulators inthe endometrium with interactions with the vascularendothelial growth factor (VEGF) system and appear toplay a major role in the regulation of blood vesselgrowth, maturation, and regression [30, 31]. In humanendometrium, ANGPT2 expression was mainly localizedto the glandular epithelium and endothelium [32]. Whileonly 11 genes upregulated in stroma (3.8-fold enrich-ment) were assigned to “vasculature development”, 116genes upregulated in complete endometria (2.4-fold en-richment) were assigned, showing that most of the genesinvolved in this process are regulated in endometrial re-gions rich in blood vessels which were not collected forthe stroma samples. These genes included severalangiogenesis-related genes such as VEGFC, VEGFD, andVEGF receptors 1, 2, and 3 as well as angiopoietin re-ceptors, and a number of endothelial cell markers. The

Table 2 Comparison of LCM RNA-seq results for endometrial localization of gene expression to results from in situ hybridization andimmunohistochemistry

Gene LE GE S Complete Other studies Technique PubMed ID

AKR1B1 5.2 5.2 2.2 5.1 LE days 12/13 P ISH 17640989, 24695626

HPGD −3.8 −1.4 −3.6 −3.0 expression in LE D12C ISH 24695626

EGFR −0.7 d d −1.0 low abundance in LE/GE D12 ISH 24012778

FGF7 2.3 1.7 d 2.1 expression in LE/GE D12P ISH 10819782

FGF9 2.2 nd nd 3.0 expression in GE D14P IHC 20393170

FGFR2 d d d d expression in LE/GE D12 ISH 28395342

ABCC1 3.5 d d 2.4 expression in LE/GE D12P (weak) ISH 27764917

LOC100738425 (ABCC4) nd 2.9 nd 2.2 expression in LE/GE D12P ISH/IHC 24695625

ABCC9 1.2 d d 2.2 expression in LE/GE D12P (weak) ISH 27764917

SLC24A4 −8.2 d −2.0 −1.9 nd in LE D12P ISH 24472379

SLCO2A1 1.0 d 2.3 d expression in LE/BV D12P ISH 24695625

SLCO4C1 1.6 d d 2.3 nd, ISH too weak, qPCR D12P up ISH 27764917

SLCO5A1 8.4 4.9 5.9 6.4 nd, ISH too weak, qPCR only D12 P ISH 27764917

GJA1 2.0 d 1.3 1.3 stroma, BV IHC 9669753

GJB1 nd nd nd −1.3 nd in LE IHC 9669753

GJB2 nd nd nd nd nd in LE IHC 9669753

IRF2 1.3 d d 1.0 cycle: low all cell types, from D12P up in LE ISH 17475929

S100G 2.9 d d 3.6 expression in LE D12P ISH 19641180

STAT2 −0.8 d d d cycle: low in all cell types, after D12P: up in stroma ISH 17475929

TRPV6 5.6 6.3 2.1 5.9 strong expression in LE, weak in GE D12P ISH 19641180

Values are log2 FC; d: expression detected; nd: expression not detectable; LE: luminal epithelium; GE: glandular epithelium; BV: blood vessels; P: pregnant; C: cyclic;ISH: in situ hybridization; IHC: immunohistochemistry

Zeng et al. BMC Genomics (2018) 19:459 Page 11 of 19

high enrichment scores obtained for “blood vessel devel-opment” and related categories is in concordance withthe finding that conceptus estrogen can increase endo-metrial vascular permeability [33].The functional category “ion transmembrane trans-

port” showed higher enrichment for genes upregulatedin stroma compared to genes upregulated in LE of preg-nant group. However, many more genes were assignedto this category for genes DE in LE. In LE and stromacalcium transporter genes had lower expression in sam-ples from pregnant gilts, particularly solute carrier family24 member 4 (SLC24A4) with a log2 FC of − 8.2. This isin agreement with a recent study where SLC24A4mRNA was detectable in LE cells during the estrouscycle by in situ hybridization, but undetectable duringearly pregnancy including Day 12 [34]. In addition, themRNAs for the calcium ion channel protein TRPV6 andthe intracellular calcium-regulatory molecule S100Gwere upregulated in all three cell types but highest in LEof pregnant gilts which is in agreement with resultsfrom quantitative PCR and localization by ISH of aprevious study [35].Altogether, the analysis of genes specifically DE in

endometrial compartments showed differential local re-sponses to the conceptus-derived signals. Selected func-tional categories important for maternal recognition ofpregnancy (MRP) and establishment of pregnancy arediscussed in the following.

Genes involved in cell communication and endometrialremodelingIn the present study, genes involved in extracellular exo-somes/vesicles (EVs) were overrepresented in DEGs ofLE, GE, and stroma. In sheep, EVs that carry genetic ma-terials (mRNAs, proteins and miRNAs) were found inuterine luminal fluid (ULF), indicating its potential rolein the conceptus-endometrial interaction [36]. The over-representation of this theme indicates a similar import-ance of EVs for embryo-maternal interactions assuggested for humans [37].In addition to cell-cell adhesion (see below), cell junc-

tions play also an important role in cell-to-cell commu-nication. Genes involved in cell junction assembly, e.g.,members of the connexin family were enriched forDEGs in LE. A number of connexins was upregulated,namely gap junction protein alpha (GJA1, GJA4, andGJA5), beta (GJB3, GJB4, and GJB5), and gamma (GJC1).Connexin 43 (GJA1) was expressed in all three cell typesand upregulated in LE and stroma. Expression of GJA4and GJB4 was only detectable in complete endometriumsamples and GJA5 only in GE. Expression and upregula-tion of GJB3 was found in all three cell types with high-est expression in LE whereas GJB5 mRNA was onlydetected in LE of Day 12 pregnant gilts. In comparison

to findings of connexin expression in human and rodentendometrium and placenta, there are some similarities,but also distinct differences probably related to thenon-invasive type of placentation in the pig. In humansas well as in rodents, expression of connexin 26 (GJB2)is suppressed in LE and connexin 43 (GJA1) is sup-pressed in stromal cells during the receptive phase andexpression of GJB2 is induced by the presence of ablastocyst and by proinflammatory factors such asPGF2a and IL1B [38]. In contrast, in porcine endomet-rium, expression of GJB2 mRNA was not detectable onDay 12 of pregnancy and the estrous cycle, respectivelyand GJA1 mRNA was upregulated on Day 12. Expres-sion of connexins in porcine endometrium could becontrolled by a combination of conceptus estrogens andIL1B2 and PGs. So far, differential expression of GJA4,GJA5, GJB3, GJB4, and GJB5 has been described only introphoblast cells of human and/or mouse placenta [39],but not in endometrial cells. In addition, several genes ofthe tight junction protein family (claudins) were foundas DE, CLDN1 upregulated in LE and stroma, CLDN3and CLDN4 upregulated only in LE, and CLDN8 down-regulated in LE and stroma. The distinct expression pat-tern of connexins and claudins in porcine endometriumsuggests an important role in the interaction of the dif-ferent functional compartments of the endometrium andwith the conceptus.The functional term “epithelial cell morphogenesis”

was significantly enriched for LE. In this context, a num-ber of growth factor and growth factor receptor geneshave been found as differentially expressed in porcineendometrium during the preimplantation phase [40].Fibroblast growth factor 7 (FGF7) mRNA was upregu-lated in LE and GE but not in stromal regions in agree-ment with the results of a previous study [41]. FGF7secreted to the uterine lumen has been suggested tohave autocrine effects on the endometrium as well asparacrine effects on the conceptus trophectoderm [42].The mRNA for the high-affinity receptor for FGF7,FGFR2 was expressed in all three endometrial cell typesat high levels, supporting FGF7 could have autocrine ef-fects on the endometrium. The mRNA of another fibro-blast growth factor, FGF9 was specifically upregulated inLE in agreement with the results in one of our previousstudies [43]. Upregulation of FGF9 expression was alsofound during the phase of maternal recognition of preg-nancy in the mare [44].In human endometrium, stroma-derived insulin-like

growth factors (IGFs) are implicated in growth regula-tion of epithelial cells [45]. In porcine endometrium,mRNAs for IGF1 and IGF2 as well as their receptorswere not DE on Day 12 but detected in all three endo-metrial cell types, with high expression levels for IGF1,IGFR1, and IGFR2. A number of mRNAs for IGF

Zeng et al. BMC Genomics (2018) 19:459 Page 12 of 19

binding proteins (IGFBP2, IGFBP3, IGFBP4, IGFBP5,IGFBP6, IGFBP7) were also detected, but only IGFBP2was found as upregulated in complete endometria, sug-gesting upregulation in blood vessel regions. Anothergrowth factor gene, EGF and its receptor (EGFR) wereDE in porcine endometrium, EGF downregulated in LEand stroma and EGFR downregulated in LE. A recentstudy in cyclic and ovariectomized gilts revealed down-regulation of endometrial EGFR expression in ovariecto-mized estradiol-treated gilts while EGF remainedunchanged [46]. Localization of expression in porcineendometrium by ISH revealed low abundance of EGFRmRNA in LE and GE between days 9 and 15 of the es-trous cycle and days 9 and 12 of pregnancy [47].The transforming growth factor beta (TGFB) signaling

pathway has been identified as an important modulatorof many endometrial functions during the sexual cycle,the implantation phase and placentation [48]. Most ofthe genes assigned to the TGFB signaling pathway werefound as DE in complete endometria and some instroma, such as bone morphogenetic protein 2 and 4(BMP2, BMP4), suggesting a role in vascular remodeling.TGFB1 and TGFB2 were expressed in LE, GE, andstroma and were upregulated in complete endometrialsamples, TGFB3 was identified as upregulated in stromaonly. Inhibin beta B subunit (INHBB) mRNA was upreg-ulated in LE and stroma.In summary, intercellular communication is regulated

by cellular junctions in a species-specific manner in con-text of the strictly non-invasive type of implantation inthe pig. Several growth factor systems are regulated inthe endometrium. The FGF system seems to be involvedin autocrine and paracrine effects on the epithelium and

conceptus, respectively. Based on the localization ofgene expression regulation of the IGF, EGFR, and TGFBsystems, suggests a role mainly in remodeling of stromalregions and blood vessels (overview in Fig. 6).

Genes related to cell adhesionMembers of the integrin family were reported to be crit-ical for endometrium-conceptus communication and im-plantation, as well as cell adhesion cascades [49]. Twointegrin alpha genes (ITGA1 and ITGAV) and six integ-rin beta genes (ITGB1, ITGB2, ITGB3, ITGB5, ITGB6,and ITGB8) showed higher mRNA concentrations incomplete endometrium on Day 12 of pregnancy. The re-sults from the LCM samples revealed four integringenes, ITGAV, ITGA3, ITGB6, and ITGB8 as upregulatedin LE, ITGB6 in addition upregulated in GE. In line withthe mRNA expression, ITGA3 protein was localized tothe uterine epithelium during early pregnancy in the pig[50]. The genes for ITGA1, ITGB1, ITGB2, ITGB3,ITGB5, ITGB6, and ITGB8 were only DE in thecomplete tissue samples. Based on the known functionsof these integrins, it is likely that they are DE in B or Tcells and/or blood vessels.The expression of osteopontin (also known as Secreted

Phosphoprotein 1, SPP1), a secreted ECM protein thatbinds to a variety of cell surface integrins, has beenshown to increase markedly in LE during theperi-implantation period in pigs [51]. In our study, SPP1mRNA was upregulated in LE and also in stroma. It isknown that expression of the integrin heterodimeralpha(v)beta(6), which is binding to SPP1, on the endo-metrium and the trophoblast is necessary for conceptus

Fig. 6 Summary of the main findings of the study. This schematic overview is based on the results of the present study of endometriallocalization of differential gene expression. Genes highlighted in red and blue color were found as up- and downregulated, respectively whencomparing pregnant to nonpregnant sows

Zeng et al. BMC Genomics (2018) 19:459 Page 13 of 19

attachment as the ability of trophectoderm cells to bindto OPN decreased after knocking down ITGAV [52].A number of further genes assigned to the GO cat-

egory “cell-cell adhesion” were upregulated in LE. Galec-tin 9 (LGALS9) mRNA was specifically upregulated inLE, and a similar expression profile was also reported inbovine endometrium [53]. Popovici et al. characterizedthe cells that express LGALS9 in human endometrium,and found that LGALS9 mRNA was significantly upregu-lated during early pregnancy in epithelial cells, whereasit was not expressed in stromal cells [54]. Additionally,syndecan 4 (SDC4) was upregulated in LE. In mice,SDC4 was found as upregulated in LE and GE, and alsoin endometrial fibroblasts [55]. Another mRNA for a celladhesion molecule, activated leukocyte cell adhesionmolecule (ALCAM) was upregulated in LE and stroma.The results of a previous study showing that ALCAM isover-expressed in human endometrial cell during theimplantation period [56] and expression of ALCAM onhuman endometrial epithelial cells and blastocysts [57]suggests that the ALCAM-ALCAM cell adhesion systemis probably involved in the interaction between the em-bryo and maternal endometrium in humans and the pig.The observed expression of genes known as in-

volved in cell adhesion, shows that the attachment ofthe conceptus trophectoderm to the LE followingafter day 13 of pregnancy is already initiated on Day12 (overview in Fig. 6).

Genes related to immune responseIn context of conceptus-endometrium interactions inthe pig, a number of signaling molecules involved inregulation of immune response have been described[58]. The pig conceptus-specific IL1B2 plays an import-ant role during conceptus elongation and establishmentof pregnancy through its effects on the uterine luminalepithelium [59, 60]. In the context of IL1B signaling,interleukin 1 receptor type 1 (IL1R1), interleukin 1 re-ceptor accessory protein (IL1RAP), and interleukin 1 re-ceptor associated kinases 3 and 4 (IRAK3, IRAK4) wereupregulated in LE in the present study. In agreementwith upregulation of IL1R1 in LE, a previous reportshowed that secretion of IL1B2 from the conceptus isassociated with increased endometrial expression ofIL1R1 [61]. In general, interleukin 1 beta is one of themajor signaling molecules in the NFKB signaling path-way. Many members of this central immune responsesignaling pathway were found to be differentiallyexpressed in LE, e.g., mRNAs for a number of NF-kappaB subunits (NFKB1, NFKB2, REL, RELB) were upregu-lated as well as NFKB inhibitors (NFKBIA, NFKBIB,NFKBIE). A recent study revealed that compared toIL1B1, the extent of NFKB activation and related geneexpression was lower in endometrium treated with

IL1B2 in the pig [59]. The lower NFKB activation ob-served for the IL1B2 secretion from the conceptus couldbe associated with the simultaneous upregulation ofNFKBIA, NFKBIB, and NFKBIE.In addition to the IL1 and NFKB signaling pathway, a

number of genes involved in the interferon type I signal-ing pathway (WP585) were obtained as DE. Members ofthe signal transducer and activator of transcription fam-ily (STAT1, STAT4, STAT5A, STAT5B) were mainlyfound as upregulated in LE, except STAT2 which wasdownregulated in LE. Protein inhibitor of activatedSTAT 1 (PIAS1) was also found as upregulated but onlyin the complete endometria. Furthermore, a number ofgenes of the type I IFN signaling pathway were differen-tially expressed. Interferon alpha and beta receptor sub-unit 1 (IFNAR1) and IFNAR2 were both upregulated inLE and GE, and all three LCM samples, respectively. Ina recent study, upregulation of IFNAR1 and IFNAR2mRNAs was also found on Day 12 of pregnancy in por-cine endometrium and positive regulation by IL1B forIFNAR1 mRNA and by IL1B and E2 for IFNAR2 mRNA[62]. In contrast, the expression of interferon regulatoryfactors IRF3, IRF4, IRF7, and IRF9 was lower in samplesfrom Day 12 pregnant gilts, IRF3, IRF4, IRF7 incomplete endometria and IRF9 in LE. Furthermore, anumber of typical IFN-stimulated genes (IFI6, IFI30,IFI44, IFI44L, IFIT1, IFIT2, IFIT3, IFITM1, IFITM3,MX1, 2′-5′-oligoadenylate synthetases (OAS1, OAS2),and radical S-adenosyl methionine domain containing 2(RSAD2) in LE and/or stroma) which are upregulated byIFNT in ruminants, showed lower expression in samplesderived from Day 12 pregnant gilts. These findings are inline with the results from our previous studies [18, 19, 63]which showed that the endometrial response to conceptussignals has very distinct characteristics in comparison ofDays 12 and 14 of pregnancy and a typical response totype I IFNs can only be observed on Day 14. In agreementwith a previous study [64], increased expression of IRF2mRNA in LE was found in pregnant endometrium sup-porting upregulation by conceptus-derived E2.However, in the pig, interferon gamma (IFNG) is the

major trophoblast-derived IFN on Day 12 of pregnancy,which is synthesized between Days 12 and 20 during thegestation period [65]. Interferon gamma receptor 2(IFNGR2) was upregulated in all three cell types of thepregnant endometrium in this study. Interferon gammaplays a crucial role in innate immune responses [66],and its functions are achieved via receptor-mediated sig-nals that lead to changes in the transcription of hun-dreds of genes [67]. But since a strong inflammatoryresponse would have negative effects on establishmentof pregnancy, IFNG effects have to be regulated. Sup-pressor of cytokine signaling 1 (SOCS1) has been identi-fied as a specific negative regulator of IFNG effects [68].

Zeng et al. BMC Genomics (2018) 19:459 Page 14 of 19

SOCS1 mRNA was upregulated in complete endometria.In addition, SOCS4 was found as upregulated in LE,whereas SOCS6 was downregulated in LE and stroma.Members of the S100 calcium binding protein A gene

family have been found as involved in innate immunity[69]. Particularly S100A7, S100A8, S100A9, and S100A12were highly upregulated in pregnant sows (log2 FC up8–12). Members of this protein family are in additionimportant for the modulation of the inflammatory re-sponse [70].Many CD molecules genes (e.g. CD14, CD34, CD36,

CD40, CD46, CD58, CD93, and CD99) were found asupregulated in complete tissue. In LCM samples, CD14was found as upregulated in GE and stroma and was notdetectable in LE. The CD14 molecule is an accessorymolecule of toll like receptor 4 (TLR4) which is import-ant for innate immune responses to bacterial and othermicrobial structures [71]. The mRNA for TLR4 wasdownregulated in LE and stroma. The expression pat-terns of CD14 and TLR4 suggest a specific and localmodulation of proinflammatory effects in the porcineduring early pregnancy. In addition, CD99 was found asupregulated in LE and stroma, which has a major regula-tory function in early T-cells [72]. Inducible T-cell costi-mulator ligand (ICOSLG) playing a role in regulation ofendometrial T-cells [73] was upregulated in LE whichcould also be involved in modulation of the maternalimmune system.The expression patterns of genes involved in immune

response reflects the effects of conceptus- andendometrium-derived signaling molecules. Overall, theinvolved genes and their spatial regulation suggests atight control of the maternal immune response to sup-port conceptus growth and to avoid negative inflamma-tory effects (overview in Fig. 6).

Genes involved in estrogen signaling and metabolismEstrogen receptor alpha (ESR1) expression has beenstudied in endometrium of gilts and sows during the es-trous cycle and early pregnancy [74–76]. No differencesin ESR1 mRNA expression was found between cyclicand pregnant endometrium on Days 11–12. Here, wefound three-fold lower mRNA concentrations in LE ofsamples from pregnant gilts. The mRNA encoding themembrane-bound G protein-coupled estrogen receptor1 (GPER1) was also downregulated, but only in thecomplete tissue samples. The fact that GPER1 was onlyfound as DE in complete endometrium samples suggestsregulation near blood vessel regions. Downregulation ofGPER1 mRNA could result from downregulation ofESR1 since it has been shown that GPER1 is regulatedby nuclear ESR1 and progesterone receptors [77].The functional categories “sterol biosynthetic process”

and “steroid metabolic process” were specifically

overrepresented for the DEGs found in stroma. Genes en-coding UDP-glucuronosyltransferases (UGTs) were upreg-ulated, UDP glucuronosyltransferase 1 family, polypeptideA6 (UGT1A6) in GE, UDP-glucuronosyltransferase 2B31(UGT2B31) in LE, and UDP-glucuronosyltransferase2B31-like (LOC100515741) in complete endometria. AsUGT enzymes are involved in glucuronidation of E2 [78],this indicates increased UGT-mediated estrogen metabol-ism on Day 12 of pregnancy. Cytochrome P450 familymembers CYP1B1, CYP7A1, CYP7B1 which are involvedin E2 synthesis and metabolism [79, 80], were also DE onDay 12 of pregnancy. For example, CYP7A1 regulatingcholesterol metabolism [81] was downregulated in LE andstroma, on the other side, CYP7B1 being related to steroidmetabolism [82] was downregulated in LE. Sulfotransfer-ase family 1E, estrogen-preferring, member 1 (SULT1E1)involved in steroid synthesis and metabolism has beenfound as upregulated in porcine endometrium duringDays 15 and 16 of pregnancy [83]. We found thatSULT1E1 is already upregulated on Day 12 in LE, GE, andstroma. Overall, the results for genes involved in E2 me-tabolism showed downregulation of genes involved in E2synthesis, while genes probably mediating inactivation ofE2 (sulfatases, UGTs) were upregulated in the endomet-rium on Day 12 of pregnancy (see overview in Fig. 6).

Genes involved in prostaglandin metabolism andsignalingMembers of the phospholipase A2 family (PLA2s), keyenzymes for the release of PG precursor molecules fromthe plasma membrane [84], were downregulated mainlyin complete endometrial tissue samples, exceptPLA2G4A which was localized to stromal areas. Thedownregulation of PLA2 genes in complete endometriacould indicate regulation associated with immune cellsor cells of blood vessels. In a recent study, a trend fordownregulation of PLA2G4A has been found on Day 12of pregnancy whereas no difference was found on Day15 [85]. The aldo-keto reductase family 1 member B(AKR1B1) is an aldose reductase enzyme that is secretedinto the extracellular spaces where it functions in thesynthesis of PGs, specifically PGF2a in the endometrium[86]. Strong upregulation of AKR1B1 mRNA was foundonly in LE and GE with strongest expression in LE,while prostaglandin-endoperoxide synthase 2 (PTGS2)was downregulated only in stroma. This leads us tohypothesize that AKR1B1, together with PTGS2 maybeimportant enzymes involved in the change of endocrineto exocrine secretion of PGF2a in porcine endometrium.Furthermore, the mRNA for HSD17B12 was upregulatedin LE. Its known function in conversion of estrone to es-tradiol is in contrast to the observed downregulation ofother genes involved in E2 synthesis, but HSD17B12 isalso known to function in arachidonic acid metabolism

Zeng et al. BMC Genomics (2018) 19:459 Page 15 of 19

thereby providing precursors for prostaglandin synthesis[87]. Together with the localization of upregulation inLE this suggests a role in PG synthesis on Day 12 ofpregnancy in porcine endometrium. The mRNA forprostaglandin E synthase (PTGES) was found as down-regulated in LE. In addition, PTGES2, PTGES3, andPTGES3L were also expressed in all three cell types(PTGES2 not in GE). The mRNA of PTGES3L was morethan 4-fold upregulated in LE of pregnant endometrium.However, the expression of PTGES3 mRNA was clearlyhigher compared to the mRNAs of the other PGE2synthases. Similar to the findings in a recent study [85],PTGES3 mRNA was slightly higher in LE on Day 12 ofpregnancy but did not reach the significance threshold(adjusted P-value 0.042 in LE, 0.021 in complete tissue).The biological effects of PGs are also controlled by PG

transport, e.g., mediated by specific transporters such astransmembrane transporters ATP-binding cassette, sub-family C, member 4 (ABCC4) and solute carrier organicanion transport family, member 2A1 (SLCO2A1). Inporcine endometrium, ABCC4 and SLCO2A1 expressionhas been found as upregulated on Day 12 of pregnancyin LE and GE, and LE and blood vessels, respectively[88]. In our study, ABCC4 was downregulated andSLCO2A1 was upregulated in LCM samples from stro-mal areas. A closer look at the RNA-seq data showedthat ABCC4 is expressed in all three cell types. Likewise,SLCO2A1 mRNA was detectable at similar expressionlevels in all three cell types. In GE, another gene(LOC100738425) similar to human ABCC4 wasexpressed and showed a 7.5-fold increase in samplesfrom Day 12 pregnant gilts. Altogether, the obtained re-sults revealed a very complex pattern of regulation forPG synthesizing, metabolizing and transporting proteins,as well as PG receptors. Basically, the results from ourendometrial LCM RNA-seq analysis suggest a specificupregulation of the PGF synthase AKR1B1 in LE andGE, specific uregulation of PG transporters in the LE,local upregulation of PGE synthases, and a generaldownregulation of PG precursor synthesis in other re-gions of the endometrium as the molecular mechanismfor the switch from endocrine to exocrine PGF2a secre-tion and the regulation of PGE2 production, whichneeds further comprehensive functional studies at theprotein level (for an overview see Fig. 6).

ConclusionsIn summary, this study comprehensively determined dif-ferential gene expression in the endometrial compart-ments/cell types LE, GE, and stroma of porcineendometrium during the preimplantation period. Incomparison to previous studies, the localization of tran-scriptomic changes in response to conceptus signals wasaddressed and used to draw a global picture of molecular

pathways involved in establishment and maintenance ofpregnancy in the pig. The findings of this study will serveas a basis for in-depth investigations of cell type-specificmolecular pathways in porcine during the phase of mater-nal recognition of pregnancy.

Additional files

Additional file 1: Table S1. Raw data statistic of RNA-seq (DOCX 15 kb)

Additional file 2: Table S2. Differentially expressed genes in luminalepithelium, glandular epithelium, stroma, and complete tissue. (XLSX1826 kb)

Additional file 3: Table S3. Selected function found in luminalepithelium, glandular epithelium, stroma, and complete tissue. (XLSX58 kb)

Additional file 4: Table S4. Top 10 differentially expressed genes inluminal epithelium (LE), glandular epithelium (GE), and stromal cells (S).(DOCX 20 kb)

Additional file 5: Table S5. Overrepresentation analysis for genes onlydifferentially expressed in three cell types and complete tissue. (XLSX687 kb)

Additional file 6: Table S6. DEGs involved in estrogen signaling,prostaglandin metabolism and signaling, and other selected pathways.(XLSX 117 kb)

AbbreviationsABCC4: ATP-binding cassette, subfamily C, member 4; ABI3BP: ABI familymember 3 binding protein; AKR1B1: Aldo-keto reductase family 1 member B;ALCAM: Activated leukocyte cell adhesion molecule; ALDH2: Aldehydedehydrogenase 2 family; ANGPT2: Angiopoietin-2; B4GALNT2: Beta-1,4-N-acetyl-galactosaminyltransferase 2; BMP: Bone morphogenetic protein;CALM: Calcium/calmodulin genes; CCDC: Coiled-coil domain containing;CDs: CD molecules; CENPK: Centromere protein K; CYP1B: Cytochrome P450family members; DE: Differentially expressed; DEGs: Differentially expressedgenes; ECM: Extracellular matrix; EGF: Epidermal growth factor;EGFR: Epidermal growth factor receptor; ESR1: Estrogen receptor alpha;ETV1: ETS variant 1; EVs: Extracellular exosomes/vesicles; FGF: Fibroblastgrowth factor; FRAS1: Extracellular matrix gene 1; FRK: Fyn related Src familytyrosine kinase; GE: Glandular epithelium; GJA: Gap junction protein alpha;GJB: Gap junction protein beta; GJC: Gap junction protein gamma;GMCSF: colony stimulating factor 2; GO: Gene ontology; GPER1: G protein-coupled estrogen receptor 1; HBE1: Hemoglobin subunit epsilon 1;HSD11B1: Hydroxysteroid 11-beta dehydrogenase 1; ICOSLG: Inducible T-cellcostimulator ligand; IGF: Insulin-like growth factor; IGFBP: Insulin-like growthfactor binding protein; IL1B2: Interleukin one beta 2; IL1R: Interleukin 1receptor; IL1RAP: Interleukin 1 receptor accessory protein; INHBB: Inhibin betaB subunit; IRAK: Interleukin 1 receptor associated kinase; IRF2BP2: Interferonregulatory factor 2 binding protein 2; ITGs: Integrin genes; LAP: Latency-associated peptide; LCM: Laser capture microdissection; LE: Luminalepithelium; LFA-1: Lymphocyte function-associated antigen-1;LGALS9: Galectin 9; LIF: Interleukin 6 family cytokine; LLC: Large latentcomplex; LTBP: Transforming growth factor beta binding protein;MMRN1: Multimerin 1; MTR: Methyltetrahydrofolate-homocysteinemethyltransferase; NCBI: National Center for Biotechnology Information;NFKB: Nuclear factor kappa-B; NPY: Neuropeptide Y; NR3C1: Glucocorticoidreceptor; OAS: 2′-5′-oligoadenylate synthetase; OPN: Osteopontin;OSTN: Osteocrin; PG: Prostaglandin; PGF2a: Prostaglandin F2α; PIAS1: Proteininhibitor of activated STAT 1; PLA2s: Phospholipase A2 family;PROX1: Prospero homeobox 1; PTGES: Prostaglandin E synthase;PTGS2: Prostaglandin-endoperoxide synthase 2; RANTES: C-C motifchemokine ligand 5; RIN: RNA integrity number; RNA-Seq: RNA-sequencing;ROCK: Rho associated coiled-coil containing protein kinase; RORB: RARrelated orphan receptor B; RSAD2: Radical S-adenosyl methionine domaincontaining 2; S100A: S100 calcium binding protein A; SALL2: Spalt liketranscription factor 2; SCN3A: Sodium voltage-gated channel alpha subunit 3;SDC4: Syndecan 4; SERPINB11: Serpin family B member 11; SLC24A4: Solute

Zeng et al. BMC Genomics (2018) 19:459 Page 16 of 19

carrier family 24 member 4; SLCO2A1: Solute carrier organic anion transportfamily, member 2A1; SLCs: Solute carrier family members; SNCA: Synucleinalpha; SOCS: Suppressor of cytokine signaling; SOSTDC1: Sclerostin domaincontaining 1; STAT: Signal transducer and activator of transcription family;SULT1E1: Sulfotransferase family 1E, estrogen-preferring, member 1;TGFB: Transforming growth factor beta; TLR4: Toll like receptor 4;TMEMs: Transmembrane protein; TNFα: Tumor necrosis factor; UGTs: UDP-glucuronosyltransferases; ULF: Uterine luminal fluid; VEGF: Vascularendothelial growth factor

AcknowledgementsWe would like to thank the Functional Genomics Center Zurich (FGCZ) forperforming Illumina sequencing of the RNA-Seq libraries and the ScientificCenter for Optical and Electron Microscopy (ScopeM) for support with per-formance of laser capture microdissection.

Availability of data and materialAll data used in this study have been included in the article and itssupplementary files. The sequence data (GSE109539) is available at NationalCenter for Biotechnology Information (NCBI) Gene Expression Omnibus(https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE109539).

FundingThis work was supported by the China Scholarship Council.

Authors’ contributionsSZ carried out the experiment, data analysis and prepared the manuscript. JBand SB developed the pipeline for data analysis. SEU provided the platformfor the laboratory work and supported the project. SB designed the study,supervised the experiments and revised the manuscript. All authors read andapproved the final manuscript.

Ethics approvalTreatments of gilts were performed in accordance with the regulations ofthe local authorities (District Government of Upper Bavaria, Veterinary Office).The performed standard procedures/treatments in animal breeding were allin accordance with the International Guiding Principles for BiomedicalResearch Involving Animals, as proposed by the Society for the Study ofReproduction, with the European Convention on Animal Experimentationand with the German Animal Welfare Act.

Consent for publicationNot applicable

Competing interestsThe authors declare that they have no competing interests.

Publisher’s NoteSpringer Nature remains neutral with regard to jurisdictional claims inpublished maps and institutional affiliations.

Author details1ETH Zurich, Animal Physiology, Institute of Agricultural Sciences, Zurich,Switzerland. 2Department for Farm Animals, University of Zurich, Geneticsand Functional Genomics, Clinic of Reproductive Medicine, Zurich,Switzerland.

Received: 5 February 2018 Accepted: 6 June 2018

References1. Guillomot M. Cellular interactions during implantation in domestic

ruminants. J Reprod Fertil. 1994;49(Suppl):39–51.2. Bazer FW, Spencer TE, Johnson GA, Burghardt RC, Wu G. Comparative

aspects of implantation. Reproduction. 2009;138(2):195–209.3. Dziuk P. Effect of migration, distribution and spacing of pig embryos on

pregnancy and fetal survival. J Reprod Fertil. 1984;33(Suppl):57–63.4. Geisert RD, Zavy MT, Moffatt RJ, Blair RM, Yellin T. Embryonic steroids

and the establishment of pregnancy in pigs. J Reprod Fertil. 1990;40(Suppl):293–305.

5. Modi DN, Bhartiya P. Physiology of embryo-endometrial cross talk. BiomedRes J. 2015;2:83–104.

6. Geisert RD, Renegar RH, Thatcher WW, Roberts RM, Bazer FW. Establishmentof pregnancy in the pig: I. Interrelationships between preimplantationdevelopment of the pig blastocyst and uterine endometrial secretions. BiolReprod. 1982;27(4):925–39.

7. Keys JL, King GJ. Microscopic examination of porcine conceptus-maternal interface between days 10 and 19 of pregnancy. Dev Dynam.1990;188(3):221–38.

8. Keys JL, King GJ. Structural changes in the luminal epithelium of theporcine uterus between days 10 and 19 of the estrous cycle. Am J Anat.1989;185(1):42–57.

9. Ford SP, Christenson RK. Blood flow to uteri of sows during the estrouscycle and early pregnancy: local effect of the conceptus on the uterineblood supply. Biol Reprod. 1979;21(3):617–24.

10. Zavy MT, Bazer FW, Thatcher WW, Wilcox CJ. A study of prostaglandin F2αas the luteolysin in swine: V comparison of prostaglandin F, progestins,estrone and estradiol in uterine flushings from pregnant and nonpregnantgilts. Prostaglandins. 1980;20(5):837–51.

11. Keys JL, King GJ. Effects of topical and systemic estrogen on morphology ofporcine uterine luminal epithelia. Biol Reprod. 1992;46(6):1165–75.

12. Ross JW, Ashworth MD, Stein DR, Couture OP, Tuggle CK, Geisert RD.Identification of differential gene expression during porcine conceptus rapidtrophoblastic elongation and attachment to uterine luminal epithelium.Physiol Genomics. 2009;36(3):140–8.

13. Hayden MS, Ghosh S. NF-κB, the first quarter-century: remarkable progressand outstanding questions. Genes Dev. 2012;26(3):203–34.

14. Ali S, Mann DA. Signal transduction via the NF-κB pathway: a targetedtreatment modality for infection, inflammation and repair. Cell BiochemFunct. 2004;22(2):67–79.

15. Almiñana C, Heath PR, Wilkinson S, Sanchez-Osorio J, Cuello C, Parrilla I, etal. Early developing pig embryos mediate their own environment in thematernal tract. PLoS One. 2012;7(3):e33625.

16. Wang Y, Hu T, Wu L, Liu X, Xue S, Lei M. Identification of non-coding andcoding RNAs in porcine endometrium. Genomics. 2017;109(1):43–50.

17. Lin H, Wang H, Wang Y, Liu C, Wang C, Guo J. Transcriptomic analysisof the porcine endometrium during embryo implantation. Genes. 2015;6(4):1330–46.

18. Samborski A, Graf A, Krebs S, Kessler B, Reichenbach M, Reichenbach H-D, etal. Transcriptome changes in the porcine endometrium during thepreattachment phase. Biol Reprod. 2013;89(6):1–16.

19. Samborski A, Graf A, Krebs S, Kessler B, Bauersachs S. Deep sequencing ofthe porcine endometrial transcriptome on day 14 of pregnancy. BiolReprod. 2013;88(4):1–13.

20. Bonnet A, Bevilacqua C, Benne F, Bodin L, Cotinot C, Liaubet L, et al.Transcriptome profiling of sheep granulosa cells and oocytes during earlyfollicular development obtained by laser capture microdissection. BMCGenomics. 2011;12(1):417.

21. Pascal LE, True LD, Campbell DS, Deutsch EW, Risk M, Coleman IM, etal. Correlation of mRNA and protein levels: cell type-specific geneexpression of cluster designation antigens in the prostate. BMCGenomics. 2008;9(1):246.

22. Niklaus AL, Pollard JW. Mining the mouse transcriptome of receptiveendometrium reveals distinct molecular signatures for the luminal andglandular epithelium. Endocrinology. 2006;147(7):3375–90.

23. Field SL, Cummings M, Orsi NM. Epithelial and stromal-specific immunepathway activation in the murine endometrium post-coitum. Reproduction.2015;150(2):127–38.

24. Brooks K, Burns GW, Moraes JGN, Spencer TE. Analysis of the uterineepithelial transcriptome and luminal fluid proteome during the peri-implantation period of pregnancy in sheep. Biol Reprod.2016;95(4):1–17.

25. Giardine B, Riemer C, Hardison RC, Burhans R, Elnitski L, Shah P, et al. Galaxy:a platform for interactive large-scale genome analysis. Genome Res. 2005;15(10):1451–5.

26. Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package fordifferential expression analysis of digital gene expression data.Bioinformatics. 2010;26(1):139–40.

27. Dennis G, Sherman BT, Hosack DA, Yang J, Gao W, Lane HC, et al. DAVID:database for annotation, visualization, and integrated discovery. GenomeBiol. 2003;4(9):R60.

Zeng et al. BMC Genomics (2018) 19:459 Page 17 of 19

28. Jeong W, Kim J, Bazer FW, Song G, Kim J. Stimulatory effects of interleukin-1 betaon development of porcine uterine epithelial cell are mediated by activation ofthe ERK1/2 MAPK cell signaling cascade. Mol Cell Endocrinol. 2016;419:225–34.

29. Jeong W, Jung S, Bazer FW, Song G, Kim J. Epidermal growth factor: porcineuterine luminal epithelial cell migratory signal during the peri-implantationperiod of pregnancy. Mol Cell Endocrinol. 2016;420:66–74.

30. Girling JE, Rogers PA. Regulation of endometrial vascular remodelling: roleof the vascular endothelial growth factor family and the angiopoietin–TIEsignalling system. Reproduction. 2009;138(6):883–93.

31. Tsuzuki T, Okada H, Cho H, Shimoi K, Miyashiro H, Yasuda K, et al. Divergentregulation of angiopoietin-1, angiopoietin-2, and vascular endothelialgrowth factor by hypoxia and female sex steroids in human endometrialstromal cells. Eur J Obstet Gyn R B. 2013;168(1):95–101.

32. Hewett P, Nijjar S, Shams M, Morgan S, Gupta J, Ahmed A. Down-regulation ofangiopoietin-1 expression in menorrhagia. Am J Pathol. 2002;160(3):773–80.

33. Geisert RD, Johnson GA, Burghardt RC. Implantation and establishment ofpregnancy in the pig. In: Geisert RD, Bazer FW, editors. Regulation ofimplantation and establishment of pregnancy in mammals. Switzerland:Springer International Publishing; 2015. p. 137–63.

34. Choi Y, Seo H, Shim J, Yoo I, Ka H. Calcium extrusion regulatory molecules:differential expression during pregnancy in the porcine uterus. DomestAnim Endocrinol. 2014;47:1–10.

35. Choi Y, Seo H, Kim M, Ka H. Dynamic expression of calcium-regulatorymolecules, TRPV6 and S100G, in the uterine endometrium duringpregnancy in pigs. Biol Reprod. 2009;81(6):1122–30.

36. Burns G, Brooks K, Wildung M, Navakanitworakul R, Christenson LK, SpencerTE. Extracellular vesicles in luminal fluid of the ovine uterus. PLoS One. 2014;9(3):e90913.

37. Greening DW, Nguyen HP, Elgass K, Simpson RJ, Salamonsen LA. Humanendometrial exosomes contain hormone-specific cargo modulatingtrophoblast adhesive capacity: insights into endometrial-embryointeractions. Biol Reprod. 2016;94(2):1–15.

38. Grummer R, Winterhager E. Blastocyst-mediated induction of endometrialconnexins: an inflammatory response? J Reprod Immunol. 2011;90(1):9–13.

39. Kibschull M, Gellhaus A, Winterhager E. Analogous and unique functions ofconnexins in mouse and human placental development. Placenta. 2008;29(10):848–54.

40. Waclawik A, Kaczmarek MM, Blitek A, Kaczynski P, Ziecik AJ. Embryo-maternal dialogue during pregnancy establishment and implantation in thepig. Mol Reprod Dev. 2017;84(9):842–55.

41. Ka H, Spencer TE, Johnson GA, Bazer FW. Keratinocyte growth factor:expression by endometrial epithelia of the porcine uterus. Biol Reprod.2000;62(6):1772–8.

42. Ka H, Jaeger LA, Johnson GA, Spencer TE, Bazer FW. Keratinocyte growthfactor is up-regulated by estrogen in the porcine uterine endometrium andfunctions in trophectoderm cell proliferation and differentiation.Endocrinology. 2001;142(6):2303–10.

43. Østrup E, Bauersachs S, Blum H, Wolf E, Hyttel P. Differential endometrialgene expression in pregnant and nonpregnant sows. Biol Reprod. 2010;83(2):277–85.

44. Merkl M, Ulbrich SE, Otzdorff C, Herbach N, Wanke R, Wolf E, et al.Microarray analysis of equine endometrium at days 8 and 12 of pregnancy.Biol Reprod. 2010;83(5):874–86.

45. Zhou J, Dsupin BA, Giudice LC, Bondy CA. Insulin-like growth factor systemgene expression in human endometrium during the menstrual cycle. J ClinEndocr Metab. 1994;79(6):1723–34.

46. Wollenhaupt K, Kettler A, Brussow KP, Schneider F, Kanitz W, Einspanier R.Regulation of the expression and bioactivation of the epidermal growthfactor receptor system by estradiol in pig oviduct and endometrium.Reprod Fertil Dev. 2001;13(2–3):167–76.

47. Jeong W, Kim J, Bazer FW, Song G. Epidermal growth factor stimulatesproliferation and migration of porcine trophectoderm cells throughprotooncogenic protein kinase 1 and extracellular-signal-regulated kinases1/2 mitogen-activated protein kinase signal transduction cascades duringearly pregnancy. Mol Cell Endocrinol. 2013;381(1–2):302–11.

48. Jones RL, Stoikos C, Findlay JK, Salamonsen LA. TGF-beta superfamily expressionand actions in the endometrium and placenta. Reproduction. 2006;132(2):217–32.

49. Kling D, Fingerle J, Harlan JM. Inhibition of leukocyte extravasation with amonoclonal antibody to CD18 during formation of experimental intimalthickening in rabbit carotid arteries. Arterioscler Thromb Vasc Biol. 1992;12(9):997–1007.

50. Bowen JA, Bazer FW, Burghardt RC. Spatial and temporal analyses ofintegrin and Muc-1 expression in porcine uterine epithelium andtrophectoderm in vivo. Biol Reprod. 1996;55(5):1098–106.

51. Garlow JE, Ka H, Johnson GA, Burghardt RC, Jaeger LA, Bazer FW. Analysis ofosteopontin at the maternal-placental interface in pigs. Biol Reprod. 2002;66(3):718–25.

52. Frank JW, Seo H, Burghardt RC, Bayless KJ, Johnson GA. ITGAV (alpha vintegrins) bind SPP1 (osteopontin) to support trophoblast cell adhesion.Reproduction. 2017;153(5):695–706.

53. Bauersachs S, Ulbrich SE, Gross K, Schmidt SE, Meyer HH, Wenigerkind H, etal. Embryo-induced transcriptome changes in bovine endometrium revealspecies-specific and common molecular markers of uterine receptivity.Reproduction. 2006;132(2):319–31.

54. Popovici RM, Krause MS, Germeyer A, Strowitzki T, von Wolff M. Galectin-9: anew endometrial epithelial marker for the mid-and late-secretory anddecidual phases in humans. J Clin Endocrinol Metab. 2005;90(11):6170–6.

55. San Martin S, Soto-Suazo M, Zorn TMT. Perlecan and Syndecan-4 in uterinetissues during the early pregnancy in mice. Am J Reprod Immunol. 2004;52(1):53–9.

56. Haouzi D, Dechaud H, Assou S, Monzo C, De Vos J, Hamamah S.Transcriptome analysis reveals dialogues between human trophectodermand endometrial cells during the implantation period. Hum Reprod. 2011;26(6):1440–9.

57. Fujiwara H, Tatsumi K, Kosaka K, Sato Y, Higuchi T, Yoshioka S, et al. Humanblastocysts and endometrial epithelial cells express activated leukocyte celladhesion molecule (ALCAM/CD166). J Clin Endocrinol Metab. 2003;88(7):3437–43.

58. Waclawik A. Novel insights into the mechanisms of pregnancyestablishment: regulation of prostaglandin synthesis and signaling in thepig. Reproduction. 2011;142(3):389–99.

59. Mathew DJ, Newsom EM, Guyton JM, Tuggle CK, Geisert RD, Lucy MC.Activation of the transcription factor nuclear factor-kappa B in uterineluminal epithelial cells by interleukin 1 Beta 2: A novel interleukin 1expressed by the elongating pig conceptus. Biol Reprod. 2015;92(4):101–13.

60. Geisert RD, Johnson GA, Burghardt RC. Implantation and establishment ofpregnancy in the pig. Adv Anat Embryol Cell Biol. 2015;216:137–63.

61. Ross JW, Ashworth MD, Mathew D, Reagan P, Ritchey JW, Hayashi K, etal. Activation of the transcription factor, nuclear factor kappa-B, duringthe estrous cycle and early pregnancy in the pig. Reprod BiolEndocrinol. 2010;8:39.

62. Jang H, Choi Y, Yoo I, Han J, Kim M, Ka H. Characterization of interferonalpha and beta receptor IFNAR1 and IFNAR2 expression and regulation inthe uterine endometrium during the estrous cycle and pregnancy in pigs.Theriogenology. 2017;88:166–73.

63. Bauersachs S, Wolf E. Uterine responses to the preattachment embryo indomestic ungulates: recognition of pregnancy and preparation forimplantation. Annu Rev Anim Biosci. 2015;3(1):489–511.

64. Joyce MM, Burghardt JR, Burghardt RC, Hooper RN, Jaeger LA, Spencer TE,et al. Pig conceptuses increase uterine interferon-regulatory factor 1 (IRF1),but restrict expression to stroma through estrogen-induced IRF2 in luminalepithelium. Biol Reprod. 2007;77(2):292–302.

65. Cencič A, Guillomot M, Koren S, La Bonnardiere C. Trophoblastic interferons:do they modulate uterine cellular markers at the time of conceptusattachment in the pig? Placenta. 2003;24(8):862–9.

66. Takaoka A, Yanai H. Interferon signalling network in innate defence. CellMicrobiol. 2006;8(6):907–22.

67. Boehm U, Guethlein L, Klamp T, Ozbek K, Schaub A, Fütterer A, et al. Twofamilies of GTPases dominate the complex cellular response to IFN-γ. JImmunol. 1998;161(12):6715–23.

68. Blumer T, Coto-Llerena M, Duong FHT, Heim MH. SOCS1 is an induciblenegative regulator of interferon lambda (IFN-lambda)-induced geneexpression in vivo. J Biol Chem. 2017;292(43):17928–38.

69. Foell D, Wittkowski H, Vogl T, Roth J. S100 proteins expressed inphagocytes: a novel group of damage-associated molecular patternmolecules. J Leukoc Biol. 2007;81(1):28–37.

70. Pruenster M, Vogl T, Roth J, Sperandio M. S100A8/A9: from basic science toclinical application. Pharmacol Ther. 2016;167:120–31.

71. Herath S, Fischer DP, Werling D, Williams EJ, Lilly ST, Dobson H, et al.Expression and function of toll-like receptor 4 in the endometrial cells ofthe uterus. Endocrinology. 2006;147(1):562–70.

72. Pettersen RD, Bernard G, Olafsen MK, Pourtein M, Lie SO. CD99 signalscaspase-independent T cell death. J Immunol. 2001;166(8):4931–42.

Zeng et al. BMC Genomics (2018) 19:459 Page 18 of 19

73. Nagamatsu T, Barrier BF, Schust DJ. The regulation of T-cell cytokineproduction by ICOS-B7H2 interactions at the human fetomaternal interface.Immunol Cell Biol. 2011;89(3):417.

74. Sukjumlong S, Persson E, Dalin AM, Janson V, Sahlin L. Messenger RNAlevels of estrogen receptors alpha and beta and progesterone receptors inthe cyclic and inseminated/early pregnant sow uterus. Anim Reprod Sci.2009;112(3–4):215–28.

75. Geisert RD, Brenner RM, Moffatt RJ, Harney JP, Yellin T, Bazer FW. Changes inoestrogen receptor protein, mRNA expression and localization in theendometrium of cyclic and pregnant gilts. Reprod Fertil Dev. 1993;5(3):247–60.

76. Sukjumlong S, Persson E, Kaeoket K, Dalin AM. Immunohistochemicalstudies on oestrogen receptor alpha (ERalpha) and the proliferative markerKi-67 in the sow uterus at oestrus and early pregnancy. Reprod DomestAnim. 2004;39(5):361–9.

77. Plante BJ, Lessey BA, Taylor RN, Wang W, Bagchi MK, Yuan L, et al. Gprotein-coupled estrogen receptor (GPER) expression in normal andabnormal endometrium. Reprod Sci. 2012;19(7):684–93.

78. Lépine J, Bernard O, Plante M, Têtu B, Pelletier G, Labrie F, et al. Specificityand regioselectivity of the conjugation of estradiol, estrone, and theircatecholestrogen and methoxyestrogen metabolites by human uridinediphospho-glucuronosyltransferases expressed in endometrium. J ClinEndocr Metab. 2004;89(10):5222–32.

79. Sasaki M, Kaneuchi M, Fujimoto S, Tanaka Y, Dahiya R. CYP1B1 gene inendometrial cancer. Mol Cell Endocrinol. 2003;202(1):171–6.

80. Sasaki M, Kaneuchi M, Sakuragi N, Dahiya R. Multiple promoters of catechol-O-methyltransferase gene are selectively inactivated by CpGhypermethylation in endometrial cancer. Cancer Res. 2003;63(12):3101–6.

81. Li T, Francl JM, Boehme S, Chiang JYL. Regulation of cholesterol and bileacid homeostasis by the cholesterol 7α-hydroxylase/steroid responseelement-binding protein 2/microRNA-33a axis in mice. Hepatology. 2013;58(3):1111–21.

82. Pettersson H, Holmberg L, Axelson M, Norlin M. CYP7B1-mediatedmetabolism of dehydroepiandrosterone and 5α-androstane-3β, 17β-diol–potential role (s) for estrogen signaling. FEBS J. 2008;275(8):1778–89.

83. Franczak A, Wojciechowicz B, Kotwica G. Transcriptomic analysis of theporcine endometrium during early pregnancy and the estrous cycle. ReprodBiol. 2013;13(3):229–37.

84. Andersen S, Sjursen W, Laegreid A, Austgulen R, Johansen B.Immunohistologic detection of non-pancreatic phospholipase A2 (type II) inhuman placenta and its possible involvement in normal parturition at term.Prostaglandins Leukot Essent Fatty Acids. 1994;51(1):19–26.

85. Seo H, Choi Y, Shim J, Yoo I, Ka H. Comprehensive analysis of prostaglandinmetabolic enzyme expression during pregnancy and the characterization ofAKR1B1 as a prostaglandin F synthase at the maternal-conceptus interfacein pigs. Biol Reprod. 2014;90(5):1–13.

86. Lacroix Pepin N, Chapdelaine P, Rodriguez Y, Tremblay JP, Fortier MA.Generation of human endometrial knockout cell lines with the CRISPR/Cas9system confirms the prostaglandin F2alpha synthase activity of aldo-ketoreductase 1B1. Mol Hum Reprod. 2014;20(7):650–63.

87. Kemilainen H, Adam M, Maki-Jouppila J, Damdimopoulou P,Damdimopoulos AE, Kere J, et al. The Hydroxysteroid (17beta)dehydrogenase family gene HSD17B12 is involved in the prostaglandinsynthesis pathway, the ovarian function, and regulation of fertility.Endocrinology. 2016;157(10):3719–30.

88. Seo H, Choi Y, Shim J, Yoo I, Ka H. Prostaglandin transporters ABCC4 andSLCO2A1 in the uterine endometrium and conceptus during pregnancy inpigs. Biol Reprod. 2014;90(5):101–10.

Zeng et al. BMC Genomics (2018) 19:459 Page 19 of 19